Women who have had a breast biopsy without evidence of cancer (benign breast disease or BBD), an estimated 1 million US women per year, are at increased risk of future breast cancer (BC). However, the majority of those with BBD, including those with high-risk histologic changes (i.e. women with atypical hyperplasia (AH) or proliferative disease without atypia (PDWA)), never experience BC. To date, efforts to combine clinical, epidemiologic and tissue based risk factors have not greatly improved discrimination of future BC at time of BBD. We recently found mammographic breast density and BBD to be independent risk factors, suggesting that tissue and image-based factors provide complementary information for BC. In addition to breast density, there are other promising mammographic features that reflect patterns, variation, texture, and composition of breast tissue, which may provide complementary, but independent, information for BC. We propose to comprehensively examine mammographic features at the level of the whole breast and in the local region of the BBD, to inform future BC risk prediction in women with BBD who are at increased risk (PDWA or AH). Specifically, we propose to Aim 1) Establish a nested case-control study of 120 incidents BC cases and 240 matched controls with a prior diagnosis of AH or PDWA and available film-mammograms at time of biopsy, within the Mayo Clinic BBD Cohort. Controls will be matched to cases on age, year and type of biopsy, overall impression (AH or PDWA) and length of follow-up. Automated mammographic features including texture, spatial variation, power spectral analysis, signal dependent noise, and features derived from kernel mappings will be assessed from the mammogram corresponding to time of biopsy; Aim 2) Identify breast-image risk factor(s) for future BC in the case-control study of women with AH or PDWA via comprehensive statistical analysis of mammographic features; and Aim 3) Evaluate breast-image risk factors from Aim 2 in an independent case-control study of 110 incident BC cases and 220 matched controls (same matching criteria as Aim 1) with prior AH or PDWA diagnosis and full-field digital mammography at time of biopsy within two large clinical practices (Mayo Clinic and Moffitt Cancer Center). As a Secondary Aim of the investigation we propose to evaluate for the first time whether mammographic features specific to the region of BBD on the mammogram associate with future risk of BC (ipsilateral vs. contralateral) using data from all 230 cases and 460 controls. Further, our experienced and productive research team has an established track record, having provided important insights in past studies on the relation of BBD, breast density and automated mammographic textures to BC risk prediction. Our proposal is highly significant, with the potential to identify novel imaging-based risk factors specific to women at above-average BC risk. Ultimately, the research may assist in targeting women that would benefit from intervention strategies at time of BBD or more aggressive breast screening for the earliest possible detection of disease.